Title :
Face recognition based on 2DPCA, DIAPCA and DIA2DPCA in DCT domain
Author :
Bengherabi, Messaoud ; Mezai, Lamia ; Harizi, Farid ; Cherie, Mohamed ; Guessoum, Abderrazak
Author_Institution :
Div. Archit. des Syst. et MultiMedia, Centre de Dev. des Technol. Av.- Algeria, Algiers
Abstract :
In this paper, we introduce 2DPCA, DiaPCA and DiaPCA+2DPCA in DCT domain for the aim of face recognition. The 2D DCT transform has been used as a preprocessing step, then 2DPCA, DiaPCA and DiaPCA+2DPCA are applied to a wtimesw upper left block of the global 2D DCT transform matrix of the original images. The experiments which are performed on the ORL face database show that: in addition to the expected considerable gain in both the training and testing time, the recognition rate is higher using 2DPCA, DiaPCA and DiaPCA+2DPCA in DCT domain than applying these three methods directly on the raw pixel images.
Keywords :
discrete cosine transforms; face recognition; principal component analysis; 2D DCT transform matrix; 2DPCA; DIAPCA; ORL face database; diagonal PCA; discrete cosine transforms; face recognition rate; image pixel; two-dimensional principal component analysis; Covariance matrix; Discrete cosine transforms; Face recognition; Feature extraction; Image databases; Multimedia systems; Performance evaluation; Performance gain; Principal component analysis; Testing; 2D Discrete Cosine Transform (2D DCT); DiaPCA+2DPCA; Diagonal PCA (DiaPCA); Principal Component Analysis (PCA); Two-Dimensional (2DPCA); face recognition;
Conference_Titel :
Systems, Signals and Devices, 2008. IEEE SSD 2008. 5th International Multi-Conference on
Conference_Location :
Amman
Print_ISBN :
978-1-4244-2205-0
Electronic_ISBN :
978-1-4244-2206-7
DOI :
10.1109/SSD.2008.4632873
